Are rubber O-rings in blocks a definite NO?

[smopoim86]

I'm working on a few waterblocks and thinking about doing a high end block for chilled.

http://www.xtremesystems.org/forums/showthread.php?t=133223

It would be similar to that, but with larger jets for the slightly thicker liquid.

I need to know if i need to make it without rubber seals and maybe advice on how to do that.

Thanks

[serialk11r]

I saw someone bring up silicone o-rings but I can't really say anything. They can withstand -75C or something like that, according to that person...

[L33T]

: ahem :

http://www.xtremesystems.org/forums/showpost.php?p=1994235&postcount=9

People have brought up the fact that under a chilling system (assuming that the computer will be turned off and on), there will be a temperature difference, leading to the expansion and contraction of the surrounding materials, including the o-ring and the two materials to be "joined" via the o-ring.

The o-ring in of itself will not be affected by temperatures commonly encountered in chilling. It is suffice to say that a 40% maximum compression on the o-ring is more than plenty.

I would recommend given a static situation, to choose a 40% compression as your design constant, and find an appropriate o-ring based on that starting point.

Silicon o-rings are indeed quite low-temperature-resistant and also high-temperature-resistant as well. I think the common ones aren't rated quite as fantastic as the lab ones -- maybe -50*C -- somewhere in there as the low limit.

Obviously I don't know what sort of temperatures you're trying to produce; I have a rather limited knowledge of building water chillers; however, I can say that for the absolute coldest temperatures (down to -65*C) would be PTFF (e.g., Teflon), which is better against chemicals such as antifreeze and other types of coolant. Silicon is a bit better as far as its ability to return to its pre-deformed shape, but PTFF is quite decent. Even alcohol would be fine to use on either of these, but stay away from something like acetone, which would ruin them. If you want something with extremely good properties across the board, but not extremely low temperature resistance (at least not in its commonly available form; if you can get excellent ones, they go down way below -200*C, which is quite enough, I daresay, for even the most Xtreme), I'd pick FEP. It is even quite good enough to use within a phase-change system, but it seems nobody does it that way -- people just join the pipes together the regular way (which makes sense).

Make sure when you get the o-ring, you remember to coat it with a lubricant! Also, the lubricant must be of a different chemical than the o-ring!

So just to summarize:
1. Pick PTFF, FEP, or silicon. I would recommend PTFF as the first choice, for this kind of project.
2. Use the right lubricant for your o-ring
3. Pick a compression. I recommend 40% (no higher!). Most dynamic projects, such as building an airmotor, allows for about 30% or lower (down to 10%) compression, but those are dynamic loads. This is a static load as far as the definition goes. I say go 40%; if something does shrink, you'll still get a good compression most likely. You'll want to calculate this! Figure out the coolant pressure, and get the pressure on the plates of the water block!

You don't want to use say a C = 30% inadvertently and then when you start up the chilling, something shrinks slightly and all of a sudden you're really getting about 10% compression. If you're at 40%, I think you'll be reasonable safe given any "shrinkage" that might occur.

Good luck!

[epion2985]

I'm working on a few waterblocks and thinking about doing a high end block for chilled.

http://www.xtremesystems.org/forums/showthread.php?t=133223

It would be similar to that, but with larger jets for the slightly thicker liquid.

I need to know if i need to make it without rubber seals and maybe advice on how to do that.

Thanks

It depends on what rubber. But the standard seals in todays blocks have been know to shrink, freeze and or crack causing leaks.

You can remedy this to an extent by using silicon based seals.

However if you want to make a block for a chiller the best way to go about it is to baze the block shut with no seals like the swiftech mcw6002.

[wdrzal]

search out severial seal and o ring manufacturers and study their line,they have o rings to fit the ticket. They make O-rings for ln2 so your temp range is no problem. you do the work though.;)

[L33T]

search out severial seal and o ring manufacturers and study their line,they have o rings to fit the ticket. They make O-rings for ln2 so your temp range is no problem. you do the work though.;)

Just use the constraints I gave you. C=40% is an excellent starting point. I assume the mating will be similar to a MCW5002 or a DD Maze4gpu... Actually I think most mating surfaces are quite similar. What you don't want is a mating join of the concentric type, such as that found in a piston-head/shaft assembly. It will be a bit more difficult to calculate, and you will not have the ability to manually "correct" a tolerance undershoot or overshoot by tightening or loosening any screws.

Use the materials that I have suggested. You will not need to weld anything shut if you calculate properly. It seems like you're building a custom water block, so you can decide what materials you use.

Start by getting two diagrams: one of the base (part touching the chip) and one of the cap (part where there's logos and the inlet and outlet). I recommend using the same material for both of the parts.

If you have a very high-pressure situation, you'll have to account for stresses/shears in different regions. Pick a few regions to do your calculations at. I'd pick the centre of the base, as well as the areas next to screws, and areas in between... It sorta depends on your setup; only after seeing a setup can you tell.

Then just calculate; you can look up the yielding stress, Poisson ratio, and elastic mod. of the material you're working with. Use a safety ratio of at least 2.0. If you want to really be on the safe side, choose something like 2.5. Find out all the stresses in each of the points.. Find the principal stresses. Get the "effective stress" of the area... and compare that value to 2.0 x yielding stress of material (or 2.5 x if that is what you picked). If sigma0 > sigmaY * 2.0, its time to increase the thickness. Sigma0 is the stress that is affecting the area; sigmaY is the yielding stress of the material. When you look up material properties, you want the yielding stress, not the ultimate stress. Ultimate stress is the point where it just goes snap! That's no good. You want the yielding stress, beyond which the material starts deforming permanently. So think of a spring. If you pull a spring apart reasonably, it'll snap back to its shape -- but keep pulling beyond a certain point, and it'll only shrink back a little bit. Keep pulling even more, and it'll snap and break.

You'll want to draw the block using SolidWorks (my favorite) or AutoCAD. Then use ANSYS to get the stresses/strains.

[epion2985]

No point in seals. Braze it shut, keep it simple, keep it safe.

PS:
solidworks love toast l33t :toast:

[L33T]

:woot: :woot: :woot: :woot: :woot: :woot: :woot: :woot: :woot: :woot:
S.................O...............L............... I...............D...............W...............O. ..............R...............K...............S



Well, there's a good thing about keeping it so that you can open it: if you get something trapped in the block, you can always open it up to washing it out. And in general, you can clean a block out very well if you can open it. Alas, the three blocks that I've used for years, when I opened them, I realized they were pristine, so I guess this is more of a "hypothetical" issue.

[smopoim86]

I drew it in autocad.... how do you do the stress testing? I'm not sure i know what you're talking about.

[L33T]

Are you familiar with ANSYS?

[smopoim86]

Are you familiar with ANSYS?
Nope.

[{.bLanK} GoD]

No point in seals. Braze it shut, keep it simple, keep it safe.

PS:
solidworks love toast l33t :toast:

I agree fully.
Safer and no doubt cheaper than expensive Teflon or silicon O-rings.

[epion2985]

I drew it in autocad.... how do you do the stress testing? I'm not sure i know what you're talking about.

various analysis are done in FEA (finite element analysis) packages. However you need to know physics and calculus extremely well to even start using them. Its not something anyone can just pick up and use. The software may calculate various things for you but you have to set parameters and model the physical situation correctly or the numbers you will get will be meaningless gibberish. Get a BS in mechanical engineering then you can start lol. Frankly thats where I am and I am still barely beginning to understand them.